Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
Role Steering of Language Models for Social SimulationsGoogle DeepMind, OpenAI · arXiv (Cornell University)Jul 9SiFAR: Synchronization-Free All-Reduce for Low-Latency LLM InferenceNVIDIA · arXiv (Cornell University)Jul 9FlowDAgger: Human-in-the-Loop Adaptation of Generative Robot Policies in Latent SpaceMicrosoft · arXiv (Cornell University)Jul 9LEXIC: Lightweight On-Device Decoding of Reading Comprehension from Eye MovementsNVIDIA · arXiv (Cornell University)Jul 9Ideas Have Genomes: Benchmarking Scientific Lineage Reasoning and Lineage-Grounded Idea GenerationMicrosoft · arXiv (Cornell University)Jul 9Understanding Axes of Difficulty For Long Context Tasks Via PredicateLongBenchNVIDIA · arXiv (Cornell University)Jul 9SQuaD-SQL: Efficient Text-to-SQL with Small Language Models via LLM-Guided Knowledge DistillationMicrosoft · arXiv (Cornell University)Jul 9SMetric: Rethink LLM Scheduling for Serving Agents with Balanced Session-centric SchedulingAlibaba · arXiv (Cornell University)Jul 9Reinforcement learning control of quantum error correctionAlphabet, Google DeepMind · Nature5 citationsJul 8CTA-Pipelining: A Latency-Oriented Spatial Scaling Method for Multi-GPU SystemsNVIDIA · arXiv (Cornell University)Jul 8Hallucination Self-Play: Bootstrapping Reinforced Detector via Evolved GeneratorMicrosoft · arXiv (Cornell University)Jul 8Is Randomness Necessary for Adaptive Data Analysis?Alphabet · arXiv (Cornell University)Jul 8MedPMC: A Systematic Framework for Scaling High-Fidelity Medical Multimodal Data for Foundation ModelsMicrosoft · arXiv (Cornell University)Jul 8From Noisy Traces to Root Causes: Structural Trajectory Analysis and Causal Extraction for Agent OptimizationMicrosoft · arXiv (Cornell University)Jul 8Decoupling Conversational Dynamics in Full-Duplex Spoken Models through Reinforcement LearningNVIDIA · arXiv (Cornell University)Jul 8Entropy Pacing Policy Optimization for Multi-Task Agentic Reinforcement LearningAlibaba · arXiv (Cornell University)Jul 8From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM AgentsAlibaba · arXiv (Cornell University)Jul 8Cambricon-QM: A Hybrid Architecture for Microscaling Format TrainingCambricon Technologies · IEEE Transactions on ComputersJul 8Quantum Thermal State Preparation for Near-Term Quantum ProcessorsAlphabet · Physical Review XJul 8Attention, not scale, drives human-AI alignment in multimodal language predictionAlphabet · npj Artificial Intelligence2 citationsJul 7How to DP-Fy Your Data: A Practical Guide to Generating Synthetic Data With Differential PrivacyAlphabet · Journal of Artificial Intelligence Research1 citationJul 7Load Balancing under Adaptive Bin DeletionsAlphabet · arXiv (Cornell University)Jul 7Compress the Cache, Not the Speech Embedding: KV Compression for Efficient Speech LLMsMicrosoft · arXiv (Cornell University)Jul 7Load Balancing under Adaptive Bin DeletionsAlphabet · arXiv (Cornell University)Jul 7MatrixFSDP: communication-free matrix optimizers under ZeRO-3 parameter shardingAlphabet · arXiv (Cornell University)Jul 7